The overall structures of human PCC and MCC holoenzymes.

(A, B) The side view (left) and top view (right) of human PCC holoenzyme (PCC-apo) (A) and MCC holoenzyme (MCC-apo) (B). Both holoenzymes are dodecamers consisting of six α subunits and six β subunits. The BC domains in one layer of MCCα subunits in the MCC-apo structure (marked by a dashed ellipse) are not resolved in the cryo-EM map, so they were docked into the structure according to the structure of BC domains in the other layer of MCCα subunits (see fig. S8). (C) Alignment of the PCCβ core in human PCC holoenzyme (PCC-apo) with that in a bacteria PCC holoenzyme (PDB code: 3N6R)(9). (D) Alignment of the MCCβ core in human MCC holoenzyme (MCC-apo) with that in a bacteria MCC holoenzyme (PDB code: 3U9T)(10). The β subunit cores in human PCC and MCC holoenzymes closely resemble their bacterial homologs, whereas the α subunits exhibit slightly different orientations in comparison to their bacterial homologs.

Domain organization of human PCC and MCC holoenzymes.

(A) Schematic illustration of the domain organization of human PCC (top) and MCC (bottom). Both PCC and MCC consist of α and β subunits. From the N-terminal to the C-terminal, each α subunit contains a BC (red), a BT (orange), and a BCCP (pink) domain. Each β subunit contains a CT domain, which can be divided into the CT-N (blue) and CT-C (yellow) subdomains. (B, C) Organization of the catalytic domains in the propionyl-CoA-bound PCC holoenzyme (PCC-PCO) (B) and MCC holoenzyme (C). (D) Alignment of the CT-N subdomain of PCCβ with that of MCCβ shows the different arrangement of the CT domains in human PCC and MCC. (E) One PCCβ subunit aligns well with the CT-N and CT-C subdomains from two adjacent MCCβ subunits in the MCC holoenzyme.

Acyl-CoA binding in human PCC holoenzyme.

(A) Surface representation of the binding pockets of biotin and propionyl-CoA in the PCC-PCO structure. (B, C) Alignment of the CT-N subdomain in the PCC-PCO structure with that in the PCC-apo structure. The conformation of the acyl-CoA binding pocket in the PCC-PCO structure is similar to that in the PCC-apo structure (B), and the AMKM motif and biotin covalently linked to it show minor conformational differences (C). The distances between the carbonyl oxygen in the ureido ring of the biotin and the main chain amide of the catalytic residues G437 and A438 in the PCC-PCO structure are 7.7 Å and 8.4 Å, respectively. (D) Alignment of the CT-N subdomain in the PCC-PCO structure with that in the ScPCC structure (PDB code: 1XNY)(15). The propionyl-CoA in the two structures has distinct binding modes. (E) Alignment of the CT-N subunit in the PCC-PCO structure with that in the MePCC structure (PDB code: 6YBP)(16). The propionyl-CoA binds at the similar positions in both structures. (F) Alignment of the CT-N subunit in the PCC-PCO structure with that in the PCC-ACO structure. The propionyl-CoA and acetyl-CoA in the two structures show highly similar binding modes.

Conformations of human MCC holoenzyme in apo and acyl-CoA-bound states.

(A– C) Alignment of the CT-N subdomain in the MCC-ACO structure with that in the MCC-apo structure. The BCCP domain moves towards the CT-C subdomain in the MCC-ACO structure compared to that in the MCC-apo structure (A). The covalently linked biotin binds to an exo-site in the MCC-apo structure but relocates to an endo-site that is closer to the catalytic residues A447 and G448 (see fig. S10) in the MCC-ACO structure (B). The helices around the acyl-CoA binding pocket show conformational differences in the MCC-ACO structure compared to that in the MCC-apo structure (C). (D, E) Surface representation of the binding pockets of biotin and acyl-CoA in the structures of MCC-apo and MCC-ACO. The acyl-CoA binding pocket adopts an open conformation in the MCC-apo structure (D) but switches to a close conformation in the MCC-ACO structure (E). (F) Alignment of the CT-N subdomain in the MCC-ACO structure with that in the PaMCC structure (PDB code: 3U9S)(10). The covalently linked biotin and its binding pocket adopt similar conformation in the two structures.

Protein purification and cryo-EM sample preparation.

(A–C) Preparation of the apo sample. The proteins eluted from Strep-Tactin®XT resin was further purified by size-exclusion chromatography (A). The fractions were analyzed by SDS-PAGE and visualized by Coomassie blue staining (B). The peak fractions were combined and analyzed by mass spectrometry, and the concentrated sample was directly frozen on a Holey Carbon filmed 300-mesh gold grid (Quantifoil) (C). (D–G) Preparation of the PCO sample. The proteins eluted from Strep-Tactin®XT resin was further purified by size-exclusion chromatography (D). The fractions were analyzed by SDS-PAGE and visualized by Coomassie blue staining (E). The peak fractions were combined and analyzed by mass spectrometry (F). The purified proteins were concentrated and incubated with propionyl-CoA, NaHCO3 and MgCl2, and mixed with ATP before frozen on a Holey Carbon filmed 300-mesh gold grid (Quantifoil) (G). (H–K) Preparation of the ACO sample. The proteins eluted from Strep-Tactin®XT resin was further purified by size-exclusion chromatography (H). The fractions were analyzed by SDS-PAGE and visualized by Coomassie blue staining (I). The peak fractions were combined and analyzed by mass spectrometry (J). The purified proteins were concentrated and incubated with acetyl-CoA, NaHCO3 and MgCl2, and mixed with ATP before frozen on a Holey Carbon filmed 300-mesh gold grid (Quantifoil) (K).

Flowchart for the cryo-EM data processing of human PCC and MCC holoenzymes.

(A) Cryo-EM data processing of the PCC-apo and MCC-apo structures. Particles from 11,481 micrographs went through multiple rounds of classification to reconstruct several cryo-EM density maps. The final cryo-EM density maps achieved the resolutions of 3.02 Å and 2.29 Å for PCC-apo and MCC-apo, respectively. For the 2D classes and 3D volumes, the red dots indicate PCC, and the blue dots indicate MCC. (B) Cryo-EM data processing of the PCC-PCO and MCC-PCO structures. Particles from 10,898 micrographs went through multiple rounds of classification to reconstruct several cryo-EM density maps. The final cryo-EM density maps achieved the resolutions of 2.80 Å and 2.36 Å for PCC-PCO and MCC-PCO, respectively. (C) Cryo-EM data processing of the PCC-ACO and MCC-ACO structures. Particles from 4,036 micrographs went through multiple rounds of classification to reconstruct several cryo-EM density maps. The final cryo-EM density maps achieved the resolutions of 3.38 Å and 2.85 Å for PCC-ACO and MCC-ACO, respectively. The scale bar in each micrograph is 50 nm.

Cryo-EM analysis of human PCC holoenzymes.

(A–D) Cryo-EM analysis of the PCC-apo structure. (A) Cryo-EM map and the corresponding Euler distribution plot of the PCC-apo structure. (B) The local resolution estimation of the PCC-apo structure. (C) The gold standard FSC curves and (D) the FSC curves for cross-validation of the PCC-apo structure. (E–H) Cryo-EM analysis of the PCC-PCO structure. (E) Cryo-EM map and the corresponding Euler distribution plot of the PCC-PCO structure. (F) The local resolution estimation of the PCC-PCO structure. (G) The gold standard FSC curves and (H) the FSC curves for cross-validation of the PCC-PCO structure. (I–L) Cryo-EM analysis of the PCC-ACO structure. (I) Cryo-EM map and the corresponding Euler distribution plot of the PCC-ACO structure. (J) The local resolution estimation of the PCC-ACO structure. (K) The gold standard FSC curves and (L) the FSC curves for cross-validation of the PCC-ACO structure. The particles were extracted with a box size of 488 pixels. The map resolution was determined by the reciprocal of the spatial frequency at FSC = 0.143, and the consistency of the structure optimization process was verified by comparison of the molecular model with the summed (the black curves) or half (the red or green curves) maps at FSC = 0.5(28, 33).

Cryo-EM analysis of human MCC holoenzymes.

(A–D) Cryo-EM analysis of the MCC-apo structure. (A) Cryo-EM map and the corresponding Euler distribution plot of the MCC-apo structure. (B) The local resolution estimation of the MCC-apo structure. (C) The gold standard FSC curves and (D) the FSC curves for cross-validation of the MCC-apo structure. (E– H) Cryo-EM analysis of the MCC-PCO structure. (E) Cryo-EM map and the corresponding Euler distribution plot of the MCC-PCO structure. (F) The local resolution estimation of the MCC-PCO structure. (G) The gold standard FSC curves and (H) the FSC curves for cross-validation of the MCC-PCO structure. (I–L) Cryo-EM analysis of the MCC-ACO structure. (I) Cryo-EM map and the corresponding Euler distribution plot of the MCC-ACO structure. (J) The local resolution estimation of the MCC-ACO structure. (K) The gold standard FSC curves and (L) the FSC curves for cross-validation of the MCC-ACO structure. The particles were extracted with a box size of 488 pixels. The map resolution was determined by the reciprocal of the spatial frequency at FSC = 0.143, and the consistency of the structure optimization process was verified by comparison of the molecular model with the summed (the black curves) or half (the red or green curves) maps at FSC = 0.5(28, 33).

The local density of human PCC and MCC holoenzymes.

(A–D) Cryo-EM maps of the BC domain (A), CT domain (B), BT domain (C) and biotinylated BCCP domain (D) in the PCC-apo structure. The biotin is covalently linked to K694 at the AMKM motif of the BCCP domain (D). (E–G) Cryo-EM maps of representative α-helices in the BC (E), CT (F) and BT (G) domains in the PCC-apo structure. (H–K) Cryo-EM maps of the BC domain (H), CT domain (I), BT domain (J) and biotinylated BCCP domain (K) in the MCC-apo structure. The biotin is covalently linked to K681 at the AMKM motif of the BCCP domain (K). (L–N) Cryo-EM maps of representative α-helices in the BC (L), CT (M) and BT (N) domains in the MCC-apo structure.

The local density of the biotinylated BCCP domains and acyl-CoA in human PCC and MCC holoenzymes.

(A–C) Cryo-EM maps of the biotinylated BCCP in the PCC-apo (A), PCC-PCO (B) and PCC-ACO (C) structures. (D–F) Cryo-EM maps of the biotinylated BCCP in the MCC-apo (D), MCC-PCO (E) and MCC-ACO (F) structures. (G–J) Cryo-EM maps of the propionyl-CoA in the PCC-PCO structure (G), the acetyl-CoA in the PCC-ACO structure (H), the propionyl-CoA in the MCC-PCO structure (I) and the acetyl-CoA in the MCC-ACO structure (J). The cryo-EM maps were displayed at an RMS level of 5 σ for (A–F) and of 4.5 σ for (G–J).

The local density of the acyl-CoA binding pocket in human PCC and MCC holoenzymes.

(A–C) Cryo-EM maps of the acyl-CoA binding pocket in the PCC-apo (A), PCC-PCO (B) and PCC-ACO (C) structures. (D–F) Cryo-EM maps of the acyl-CoA binding pocket in the MCC-apo (D), PCC-PCO (E) and PCC-ACO (F) structures. All cryo-EM maps were displayed at an RMS level of 4.5 σ.

The Cryo-EM maps of human PCC and MCC holoenzymes.

(A–C) Cryo-EM maps for the PCC-apo (RMS Level = 2.8 σ) (A), PCC-PCO (RMS Level = 2.7 σ) (B) and PCC-ACO (RMS Level = 3.5 σ) (C). The cryo-EM maps are displayed in semitransparent white color and superimposed with the corresponding stereo models. The density of one PCCα subunit in the PCC-ACO structure is partially missing (indicated by dashed circle in panel C). (D–F) Cryo-EM maps for the MCC-apo (RMS Level = 2.3 σ) (D), MCC-PCO (RMS Level = 2.0 σ) (E) and MCC-ACO (RMS Level= 2.8 σ) (F). The density of three MCCα subunits in each MCC holoenzyme structure are partially missing (indicated by dashed circles in panels D–F).

Comparison of the domain structures of human PCC and MCC with that of their bacteria homologs.

(A–C) Alignments of the BC domain (A), BT domain (B) and biotinylated BCCP domain (C) in human PCC structure (PCC-apo, colored) with that in a bacteria PCC structure (PDB code: 3N6R, gray)(9). (D–F) Alignments of the BC domain (D), BT domain (E) and biotinylated BCCP domain (F) in human MCC structure (MCC-apo, colored) with that in a bacteria MCC structure (PDB code: 3U9T, gray)(10).

The biotin binding sites in the structures of human MCC holoenzyme.

(A) The covalently linked biotin in the MCC-apo structure binds to a site distant from the catalytic residues in the CT domain. This site is thus referred to as the exo-site. The distances between the carbonyl oxygen of the ureido ring of biotin and the main chain nitrogen atoms of the catalytic residues A447 and G448 in the CT domain are 7.8 and 8.1, respectively. (B) The covalently linked biotin in the MCC-ACO structure moves to a site closer to the catalytic residues A447 and G448 in the CT domain. This site is referred to as the endo-site. The distances between the carbonyl oxygen of the ureido ring of biotin and the main chain nitrogen atoms of the catalytic residues A447 and G448 in the CT domain are shortened to 3.6 and 3.5, respectively.

Comparison of human MCC holoenzyme structures in complex with acetyl-CoA and 3-methylcrotonyl-CoA.

(A) Surface representation of the binding pocket of biotin and acyl-CoA in human MCC holoenzyme structure in complex with its natural substrate, 3-methylcrotonyl-CoA (PDB code: 8J4Z). The acyl-CoA binding pocket adopts a closed conformation, which is similar to that in the structure of acetyl-CoA-bound MCC (refer to Fig. 4E). (B) Alignment of the CT-N subdomain in the MCC-ACO structure with that in the 3-methylcrotonyl-CoA-bound MCC (MCC-MCO) structure. The biotin molecules in both structures bind to an endo-site, and the biotin binding pockets adopt a similar conformation. (C) Alignment of the CT-N subunit in the MCC-ACO structure with that in the 3-methylcrotonyl-CoA-bound MCC (MCC-MCO) structure.

Cryo-EM data collection, reconstruction, refinement, and validation statistics of human PCC structures.

Cryo-EM data collection, reconstruction, refinement, and validation statistics of human MCC structures.